Packet loss and increasingly high latency at fixed intervals

[MicrosoftSam]

Hello Bell community !

4-5 days ago I started experiencing weird latency issues. Here you can find a log of my ping output over a period of time to better understand the issue. Every so often the latency gets really high then slowly drops to normal (5ms to 8ms) then it jumps up again. Of course the modem integrated speed test tool is unable to detect such problem as it does not do a continous ping and I de believe that their speed test server is inside Bell's network which is not real world metrics IMO. Because of that Bell's technical support tells me that they are unable to help as the integrated speed test does not show any problems.

My current setup is like this at the moment : Windows 11 Machine -> Intel PCI-E Dual SFP+ (LC Fiber) -> TL-SG3428X SFP+ -> Cisco SFP+ DAC cable -> Sophos SG 230 (pfSense) SFP+ -> 10G SFP+ to RJ45 module -> CAT6a STP cable -> 10G port on Giga Hub

I'm usually doing PPPoE straight in pfSense like this : 

• No VLAN tagging
• MTU : default
• MSS : default

For testing I did the followings : 

• Tried doing PPPoE using only RJ45 from Sophos box to Giga Hub
  • Tried Giga Hub 1G and 10G ports
• Tried connecting my machine to the switch using the builtin RJ45 interface
• Tried connecting my machine to the Sophos box directly
• Changed the Sophos SG 230 for a Sophos SG 450 (same pfSense config)
  • This determined it is not a hardware issue on my side
    • Inter-VLAN traffic does not experience the same latency issue
• Tried changing MTU to 1492 and MSS to 1452 for the PPPoE connection
• Tried using Advanced DMZ instead
• Tried doing simple double NAT
• Tried using Sophos XG v19 as an OS instead (Yes I actually did put in a new drive an install and configure it -_-)
• Replaced cabling
• Tried connecting directly into the Giga Hub (read below)

The issue is also the same at a greatly lesser extent if plugged into the Giga Hub directly but I figure this is because there is actually no load on it since everything at home is currently in VLANs but I experience the same spikes (70ms to 100ms). For exemple, doing a speed test on speedtest.net will often cause packet loss and/or high latency spikes due to load. Giga Hub is on version 1.15.2 but the issue was present on v1.13.

I have never had this problem in 4 years with the exact same setup. (I do internal monitoring with Prometheus -> Grafana) I previously had the HH3K before upgrading to 1.5Gbps but also never had the issue with the Giga Hub until now. This is a show stopper for me, worse latency then hybrid fiber-coax.

I am wondering if someone here has a suggestion to further troubleshoot the problem or maybe someone working at Bell can help. @BellPatricia Maybe ?

Thank you for reading this and I am looking forward to read the replies.

[User4802]

Thanks for the reply,

here are some image of the tests. Taken at around 9:45 pm yesterday. Not the worst result I got yesterday, but the only image I have on my phone. 

ping : 

 traceroute : 

 

[dks]

From the traceroute your highest pings are in your last connection. All the other pings are within expectation and after the first appear to be outside Bell’s system. While I appreciate your frustration, there isn’t much we can do. If you could post a traceroute to your gaming server, we might make more comments. 

[AndreM]

I've been experiencing intermittent latency on my Bell Fibe 3.0Gbps. I'm hoping someone will have knowledge or advice on how to address this.

On Oct 3rd, a Bell technician came out to my home to deal with a different issue (frequent error 1201 disconnects). The technician replaced the HomeHub 4000, and said he also replaced the fibre connection in the junction box outside my home. Although this fixed the original problem, it's also when the intermittent latency spikes started.

• The latency spikes tend to be in the 50ms-150ms range.
• I've observed latency spikes as short as a few seconds, and as long as 5 minutes.
• All testing is done with a device connected directly to the HH4000 by ethernet.
• I can't discern any pattern in the timing of the latency spikes.
• I've tried running the Virtual Repair Tool. I didn't resolve the issue.

I've done traceroutes during latency spikes and I've attached some screenshots. Those high latency hops (42ms, 138ms, 63ms) are all Bell IPs, I believe.

Any suggestions?

[Vanadiel]

Routers tend to give timeouts and echo response handling lower priority than packet forwarding. This allows the switch to operate more efficiently as forwarding is much simpler to process than generating a new ICMP responce. As a result busier switches will take longer to generate the response.

Some switches may have ICMP generation disabled which makes diagnosing network issues with them diffficult. Other may occasionally get busy enough that they fail to respond.

If you use a tool which displays the IP address of the routers along the way, you will usually see the switches which are slower are core switches in major cities.

What you are seeing can be considered normal and is not necessarily an issue. As long as the time between you and the destination is low, that should generally be the most important metric.